PrimeGrid Races 2017

[Ken g6]

Current race: PPS Sieve, December 18-21 (16:28 UTC)

Welcome to another year of PrimeGrid races! Here's the (tentative) list of all races for this year:

What you need:

• One or more fast x86 processors. (Even slow ones might do!)
• Windows (Vista or later 64-bit, or XP or later 32-bit), Linux, or MacOS 10.4+.
• BOINC, attached to PrimeGrid (http://www.primegrid.com/).
• Your PrimeGrid Preferences with only the above project(s) selected in the Projects section.

What may help LLR (the central 6 races):

• An Intel Sandy Bridge or later ("Core series" other than first-generation) processor with AVX may be 20-70% faster than with the default application. Sadly, that does not include Pentium or Celeron processors, or AMD processors.
• Turning off hyper-threading. It really does help, even more than just doing half as much work as you have virtual cores.
• Faster RAM might help on many races, as long as it's stable.
• In certain races, usually those with longer WUs, it may help to enable multi-core processing with app_config.xml.
• Juggling in some extra WUs may help in all races.

What may help in other races:

• A GPU helps in the last two races.
• Turning on hyper-threading may help.
  
• A 64-bit OS may help.

What won't help (but won't hurt either):

• A large amount of RAM.
• Any Android devices.

What won't help (and will hurt, sort of):

• Unstable processors. (Invalid work will be deducted! If Prime95 worked recently on your processor, it should be stable.)
• Work not downloaded and uploaded within the challenge. (It's not counted.) Should you not be able to be in front of one or more computers at that time, there are several options:
  • You can often set BOINC's network connection preferences to wait until a minute or two after race time.
  • And for short work units, you can just set the queue level very low (0.01 days). This also makes it more likely that you will be a prime finder rather than a double-checker. But you might want to raise their queue size after the challenge is underway.

Welcome and good luck to all!

P.S. If no one has posted stats lately, try tracking your stats with my user script. With that installed, visit the current race's Team stats link for TeAm stats.

 

[TennesseeTony]

Only two GPU races?

And sheesh, three LONG races. At least the others are short.

 

[Ken g6]

There are several large challenges this year, but there's a reason: There's three projects we wish to promote heavily this year: GCW Sieve, PSP, and SoB. All three challenges are long because we want to give them the biggest possible push. Also, even though the tasks won't be as large as they normally are, PSP and SoB will still be very large tasks and need long durations so that people with non-AVX CPUs can participate.

I just noticed the start times vary a lot, too, for no apparent reason.

 

[petrusbroder]

Thanks a lot for the schedule. I don't know if I'll join, but the schedule makes it easier to plan.

 

[StefanR5R]

After the December F@H adventure I planned to take my two dual-Xeon boxes off of DC for a while. But not having received the December power bill yet, and feeling kind of indebted to this Newton guy, I am on the edge. Or possibly already over the edge.

 

[Ken g6]

After the December F@H adventure I planned to take my two dual-Xeon boxes off of DC for a while. But not having received the December power bill yet, and feeling kind of indebted to this Newton guy, I am on the edge. Or possibly already over the edge.

You know you're gonna have to face it, you're addicted to love DC.

 

[StefanR5R]

You know you're gonna have to face it, you're addicted to love DC.

I feel simultaneously encouraged and frightened.

 

[StefanR5R]

I checked a dual-socket Broadwell-EP for the effect of hyperthreading on GCW Sieve:

HT off: 3705 s runtime, 3702 s CPU time per task (average of 28 tasks)
HT on: 7237 s runtime, 7230 s CPU time per task (average of 56 tasks)
hence, improvement of throughput with HT on: 2.4 %​

This was "Generalized Cullen/Woodall (Sieve) v1.00" running on E5-2690 v4 @ 3.2 GHz.

 

[Ken g6]

And we're off!

 

[Orange Kid]

And we're off!

Yes we are

 

[TennesseeTony]

Regrettably, my contribution to the CPU races will be limited. I and my CPUs are committed to fighting the Zika virus on WCG, as I have many family members and a daughter of child bearing age.

I only have 3 of 8 cores to offer for a short time, on my primary Milkyway cruncher. (Which I hope to retire before this ten day race is up, pending me finding spare time to build Skeletor.)

 

[Ken g6]

Day 1 stats:

Rank___Credits____Username
11_____981236_____xii5ku
159____81887______Orange Kid
171____72621______Ken_g6
183____66081______waffleironhead
216____50255______GLeeM
252____38218______geecee
418____3282___10esseeTony

Rank__Credits____Team
6_____2014185____Duke University
7_____1525923____[H]ard|OCP
8_____1487447____Rechenkraft.net
9_____1293583____TeAm AnandTech
10____1248511____AMD Users
11____1199660____BOINC@Poland
12____1052341____Christians

Tony's not doing much, but xii5ku (Stefan?) is making up for that pretty well. It looks like sieves do well with more cores thrown at them, almost regardless of the core's performance. So let's throw more cores at the problem!

 

[StefanR5R]

Yep, xii5ku = StefanR5R. I am surprised to have landed at such a high individual rank, given that there are a bunch of participants who commandeer a whole fleet of DC nodes, whereas I just have my two dual-socket machines (and 4-5 normal PCs, depending on office hours and, as I have just noticed, depending on the occasional unexplained internet outage at the office).

 

[Ken g6]

That looks like a fleet to me, Stefan! I have 4 machines going, totaling 10 real cores. I might be able to add a few more.

 

[Ken g6]

Day 2 stats:

Rank___Credits____Username
12_____2006347____xii5ku
77_____386684___10esseeTony
167____177458_____Orange Kid
189____150720_____Ken_g6
204____133251_____waffleironhead
224____117411_____geecee
227____113036_____GLeeM
382____27304______zzuupp

Rank__Credits____Team
7_____4897185____[H]ard|OCP
8_____3598521____Duke University
9_____3338269____Rechenkraft.net
10____3112214____TeAm AnandTech
11____2846270____BOINC@Poland
12____2778230____AMD Users
13____2365779____Christians

Looks like Tony gave us the boost we needed to get into 10th.

I also updated my user script linked from the first post to support https.

 

[TennesseeTony]

I'll put about another days worth into it, if I can stand losing, lol, then back to WCG.

 

[StefanR5R]

I checked a dual-socket Broadwell-EP for the effect of hyperthreading on GCW Sieve:

[...] improvement of throughput with HT on: 2.4 %​

This was a 2x quadchannel DDR4 system. I now also checked a 1P singlechannel DDR3 system with an i7-4900MQ (mobile Haswell with 4C/8T):

HT on, 4 concurrent tasks, 3.3 GHz average core frequency:
3997 s runtime, 3996 s CPU time per task (average of 20 tasks)
HT on, 8 concurrent tasks, 3.15 GHz average core frequency:
7123 s runtime, 7112 s CPU time per task (average of 20 tasks)
hence, improvement of throughput with hyperthreading: 12 %​

Surprisingly, the HT benefit is even better than on the Broadwell-EP, even though this mobile CPU clocks down a little bit when subjected to 8 concurrent tasks.
I did not bother to check whether I can switch HT off in the BIOS, AFAIR I can't do so on this laptop.

 

[Orange Kid]

Thanks for the updates.
Tony you know you wanna, you know you wanna, you know you wanna

trying a little forum hypnotics, probably won't work.

 

[StefanR5R]

Perhaps my 2P boxes will still be available for DC for a certain time after the Newton Challenge is over. If so, then I'll see if I can produce as many WCG OpenZika credits as TennesseeTony's GCW Sieve credits from the Newton Challenge (converted 7:1 of course).

 

[TennesseeTony]

 

[Ken g6]

Day 3 stats:

Rank___Credits____Username
12_____3126985____xii5ku
50_____976503___10esseeTony
178____273037_____Orange Kid
184____257759_____geecee
201____234275_____Ken_g6
226____194422_____waffleironhead
238____176936_____GLeeM
310____105406_____zzuupp

Rank__Credits____Team
6_____13602435___Gridcoin
7_____8978451____[H]ard|OCP
8_____5378034____Rechenkraft.net
9_____5345326____TeAm AnandTech
10____4651231____BOINC@Poland
11____4416119____AMD Users
12____3852378____Christians

Geecee brought some more power to the party and passed me. He might be about to pass Orange Kid too!

And we're really close to Rechenkraft.net.

 

[geecee]

Thanks for the stats Ken. Since it's winter, I threw all my boxes into it starting yesterday. So you've got my motley assortment of 12 old sandy/ivy bridge cores, and 2 haswells. Only box I have left is an old X6 box that I haven't booted up in a while. Would it be of any use? My kvm and spare monitor are all connected, so I'll probably have to hook that up to a TV or something lol.

 

[TennesseeTony]

Thanks for the organization and stats, Ken!

Perhaps my 2P boxes will still be available for DC for a certain time after the Newton Challenge is over. If so, then I'll see if I can produce as many WCG OpenZika credits as TennesseeTony's GCW Sieve credits from the Newton Challenge (converted 7:1 of course).

I am humbled that you would contribute to fighting the Zika virus so that I may participate in the PrimeGrid challenge! I don't know what to say except...Thanks!

I am assuming this indicates that you'd like me to continue with PrimeGrid. Be aware though that PrimeGrid awards far more points than WCG, and even your fleet would produce perhaps only 80,000 (after the division) points per day running WCG. So perhaps a time based contribution is more appropriate.

That said, I hope you get some work for your rigs! Good to have some income I imagine.

 

[StefanR5R]

Day 3 stats:

Nice... TeAm's credits/day:

day 1: 1.3 M
day 2: 1.8 M
day 3: 2.2 M​

According to the challenge forum thread, the percentage of validated work of all returned work was globally:

day 1: 51 %
day 2: 64 %
day 3: 73 %​

Assuming that percentage is also true for the TeAm's returned tasks, then our daily sum of pending + validated credits/day would be:

day 1: 2.5 M
day 2: 2.8 M
day 3: 3.1 M​

i.e. 10 % ramp up from day to day so far, if this extrapolation isn't totally off.

Duke University dropped from 8th to 13th, I hope they are alright... http://emergency.duke.edu/

I am assuming this indicates that you'd like me to continue with PrimeGrid.

Well, I leave it entirely to your judgment, and to Orange Kid's psychic powers. I was just excited about the TeAm's figures on day 2.

Be aware though that PrimeGrid awards far more points than WCG, and even your fleet would produce perhaps only 80,000 (after the division) points per day running WCG. So perhaps a time based contribution is more appropriate.

Whoops, I wasn't aware of that. I'll see how that pans out. Now that you pointed it out - I actually did a brief warm-up with WCG (all subprojects mixed) shortly before the PG race, and the peak day had 318,000 WCG-credits/day for 56 days CPU-time/day.
(edit)
Looking at stats.free-dc.org, Jan 6 of 10esseeTony at PG is 9 times worth of Jan 3 of xii5ku at WCG. Credit for credit, it'll be tough to match indeed.

 

[Orange Kid]

Thanks for the stats

An X6.........no, no ,no that's no good, especially if you hook it up to a TV. Oh wait that's what's running in the family room.